<![CDATA[Abstract. Autoregressive Fractionally Integrated Moving Average (ARFIMA) is capable of describing both short and long-memory time series through the use of fractional differencing (d) values. This study aims to apply the ARFIMA (p,d,q) model to predict the frequency of earthquakes in west of Java, Indonesia, in upcoming periods. Utilizing secondary data from United States Geological Survey (USGS) spanning from 1971 to 2023, the parameters (p,q) were estimated using the maximum likelihood estimation method, while the differencing parameter (d) was estimated using the Rescaled Range Statistics (R/S) method, resulting in d = 0,273. The best fit model was ARFIMA (1;d;1) with the equation (1-〖〖∅_1 B)(1-B)〗^0,273 Z〗_t=θ_1 (B) e_t and with an AIC value of 110,883. The model predicts 7 future periods, indicating a general increase in earthquake activity in west of Java, although fluctuations in the predictions suggest a tendency towards decreasing volatility.Abstrak. Autoregressive Fractionally Integrated Moving Average (ARFIMA) mampu menjelaskan runtun waktu jangka pendek (short memory) maupun jangka panjang (long memory) dengan nilai differencing (d) bernilai pecahan. Tujuan utama penelitian ini adalah bagaimana penerapan model ARFIMA (p,d,q) dalam memprediksi banyak gempa bumi di barat Pulau Jawa pada periode selanjutnya. Menggunakan data sekunder USGS (United States Geological Survey) tahun 1971-2023, estimasi parameter (p,q) menggunakan metode maximum likelihood d dan estimasi parameter differencing (d) dengan metode analisis Rescaled Range Statistics (R/S) memberikan hasil d=0,273, dimana model terbaik terpilih adalah ARFIMA(1;d;1) dengan persamaan model (1-〖〖∅_1 B)(1-B)〗^0,273 Z〗_t=θ_1 (B) e_t dan nilai AIC sebesar 110,883 yang menghasilkan 7 periode prediksi dengan pergerakan kejadian gempa bumi di barat Pulau Jawa relatif meningkat meskipun fluktuasi prediksi cenderung menurun.]]>
